I have limited stat/coding knowledge yet I try to do user clustering using unsupervised method using R.

I have about 2795 observations gained from survey (mixture of categorical and scale questions).

Our purpose is to cluster users in distinctive groups. We've used the questionnaire proven to be effective in its industry.

At first we turned the data as binary format and tried to use EM-algorithm to cluster but the BIC was unreasonable. Through research I've found that trying MCA would be more suitable for this analysis and factorized the dataset for it.

The result seems still vague and I am unable to do the clustering it seems. My assumption is that either we have very similar type of users or the variables are just too many or the data observations are just not enough.

In what approach should I tab into this problem to create distinctive clusters from this results? Any type of tips would be appreciated. Thanks!


put on hold as too broad by gung Aug 19 at 18:09

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • $\begingroup$ The links are currently unreadable. In general, "... what approach should I [try]...", and "Any... tips...", are indications of a question that is too broad to be answered well in a context like this. $\endgroup$ – gung Aug 19 at 18:09

You seem to have misinterpreted the BIC. It can never really be "unreasonable". It is a metric that allows you to compare different numbers of clusters. The solution with the smallest BIC is the one you should start with, checking to see if the results make sense for your application.

Doing MCA prior to clustering is not generally advisable, as you lose a lot of information when you do the MCA, and, you transform the space. It can be useful, but I'd suggest you try harder without it first.


Not the answer you're looking for? Browse other questions tagged or ask your own question.